ARML Competition 2009

Paul J. Karafiol, Head Writer

Paul Dreyer
Edward Early
Zuming Feng
Zachary Franco
Chris Jeuell
Andy Niedermaier
Leo Schneider
Andy Soffer
Eric Wepsic
Walter Carlip
May 30-31, 2009

Individual Problems

Problem 1. Let p be a prime number. If p years ago, the ages of three children formed a geometric sequence with
a sum of p and a common ratio of 2, compute the sum of the childrens current ages.
Problem 2. Define a reverse prime to be a positive integer N such that when the digits of N are read in reverse
order, the resulting number is a prime. For example, the numbers 5, 16, and 110 are all reverse primes.
Compute the largest two-digit integer N such that the numbers N, 4 N , and 5 N are all reverse primes.
Problem 3. Some students in a gym class are wearing blue jerseys, and the rest are wearing red jerseys. There are
exactly 25 ways to pick a team of three players that includes at least one player wearing each color. Compute
the number of students in the class.
Problem 4. Point P is on the hypotenuse EN of right triangle BEN such that BP bisects EBN . Perpendiculars
1
1
P R and P S are drawn to sides BE and BN , respectively. If EN = 221 and P R = 60, compute BE
+ BN
.
Problem 5.

Compute all real values of x such that log2 (log2 x) = log4 (log4 x).

Problem 6. Let k be the least common multiple of the numbers in the set S = {1, 2, . . . , 30}. Determine the
number of positive integer divisors of k that are divisible by exactly 28 of the numbers in the set S.
Problem 7. Let A and B be digits from the set {0, 1, 2, . . . , 9}. Let r be the two-digit integer A B and let s be
the two-digit integer B A, so that r and s are members of the set {00, 01, . . . , 99}. Compute the number of
ordered pairs (A, B) such that |r s| = k 2 for some integer k.
Problem 8. For k 3, we define an ordered k-tuple of real numbers (x1 , x2 , . . . , xk ) to be special if, for every i
such that 1 i k, the product x1 x2 . . . xk = x2i . Compute the smallest value of k such that there are at
least 2009 distinct special k-tuples.
Problem 9.
Problem 10.

A cylinder with radius r and height h has volume 1 and total surface area 12. Compute
If 6 tan1 x + 4 tan1 (3x) = , compute x2 .

1
r

+ h1 .

Individual Answers

Answer 1. 28
Answer 2. 79
Answer 3. 7
1
60

Answer 4.
Answer 5.

Answer 6. 23
Answer 7. 42
Answer 8. 12
Answer 9. 6
Answer 10.

158 3
33

Individual Solutions

Problem 1. Let p be a prime number. If p years ago, the ages of three children formed a geometric sequence with
a sum of p and a common ratio of 2, compute the sum of the childrens current ages.
Solution 1. Let x, 2x, and 4x be the ages of the children p years ago. Then x + 2x + 4x = p, so 7x = p. Since p
is prime, x = 1. The sum of the childrens current ages is therefore (1 + 7) + (2 + 7) + (4 + 7) = 28.
Problem 2. Define a reverse prime to be a positive integer N such that when the digits of N are read in reverse
order, the resulting number is a prime. For example, the numbers 5, 16, and 110 are all reverse primes.
Compute the largest two-digit integer N such that the numbers N, 4 N , and 5 N are all reverse primes.
Solution 2. Because N < 100, 5 N < 500. Since no primes end in 4, it follows that 5 N < 400, hence
N 79. The reverses of 5 79 = 395, 4 79 = 316, and 79 are 593, 613, and 97, respectively. All three of these
numbers are prime, thus 79 is the largest two-digit integer N for which N , 4 N , and 5 N are all reverse primes.

Problem 3. Some students in a gym class are wearing blue jerseys, and the rest are wearing red jerseys. There are
exactly 25 ways to pick a team of three players that includes at least one player wearing each color. Compute
the number of students in the class.
Solution 3. Let r and b be the number of students wearing red and blue jerseys, respectively. Then either we
choose two blues and one red or one blue and two reds. Thus
     
b
r
b
r
+
= 25
2 1
1 2

rb(b 1) br(r 1)
+
= 25
2
2
rb((r 1) + (b 1)) = 50

rb(r + b 2) = 50.
Now because r, b, and r + b 2 are positive integer divisors of 50, and r, b 2, we have only a few possibilities
to check. If r = 2, then b2 = 25, so b = 5; the case r = 5 is symmetric. If r = 10, then b(b + 8) = 5, which is
impossible. If r = 25, then b(b + 23) = 2, which is also impossible. So {r, b} = {2, 5}, and r + b = 7.
Problem 4. Point P is on the hypotenuse EN of right triangle BEN such that BP bisects EBN . Perpendiculars
1
1
P R and P S are drawn to sides BE and BN , respectively. If EN = 221 and P R = 60, compute BE
+ BN
.
1
1
Solution 4. We observe that BE
+ BN
= BE+BN
BEBN . The product in the denominator suggests that we compare
areas. Let [BEN ] denote the area of 4BEN . Then [BEN ] = 12 BE BN , but because P R = P S = 60, we
can also write [BEN ] = [BEP ] + [BN P ] = 21 60 BE + 12 60 BN . Therefore BE BN = 60(BE + BN ), so
1
1
BE+BN
1
BE + BN = BEBN = 60 . Note that this value does not depend on the length of the hypotenuse EN ; for a
1
1
+ BN
is invariant.
given location of point P , BE
PS
BE
BE60
Alternate Solution: Using similar triangles, we have ER
=
P R = SN = BN , so
60
1
1
1
2
2
2
BE + BN = 221 . Using algebra, we find that BE = 204, BN = 85, and 204 + 85 = 60 .

60
BN 60

BE
BN

and

Problem 5.

Compute all real values of x such that log2 (log2 x) = log4 (log4 x).
y

Solution 5. If y = loga (loga x), then aa = x. Let y = log2 (log2 x) = log4 (log4 x). Then 22 = 44 = (22 )(2

2y+1
22
, so 2y + 1 = y, y = 1, and x = 2.
(This problem is based on one submitted by ARML alum James Albrecht, 1986-2007.)

2 y

Alternate Solution: Raise 4 (or 22 ) to the power of both sides to get (log2 x)2 = log4 x. By the change of

x)2
log x
log 2
1/2
= 2.
base formula, (log
(log 2)2 = 2 log 2 , so log x = 2 , thus x = 2

Alternate Solution: Let x = 4a . The equation then becomes log2 (2a) =

log4 a. Raising 4 to the power of
each side, we get 4a2 = a. Since a 6= 0, we get 4a = 1, thus a = 41 and x = 2.
Problem 6. Let k be the least common multiple of the numbers in the set S = {1, 2, . . . , 30}. Determine the
number of positive integer divisors of k that are divisible by exactly 28 of the numbers in the set S.
Solution 6.
know that k = 24 33 52 7 11 13 17 19 23 29. It is not difficult to see that the set
 k We
k k k
k
k
k
T1 = 2 , 3 , 5 , 17 , 19
, 23
, 29
comprises all divisorsnof k that are divisible by exactly 29 of the numbers
o in the
k
set S. Let P = {2, 3, 5, 17, 19, 23, 29}. Then T2 = p1 p2 , where p1 and p2 are distinct elements of P consists


of divisors of k that are divisible by exactly 28 of the numbers in the set S. There are 72 = 21 elements in T2 .
Furthermore, note that k7 is only divisible by 26 of the numbers in S (since it is not divisible by 7, 14, 21, or
k
k
28) while 11
and 13
are each divisible by 28 of the numbers in S. We can also rule out k4 (27 divisors: all but
k
k
8, 16, and 24), 9 (27 divisors), 25
(24 divisors), and all other numbers, thus the answer is 21 + 2 = 23.
Problem 7. Let A and B be digits from the set {0, 1, 2, . . . , 9}. Let r be the two-digit integer A B and let s be
the two-digit integer B A, so that r and s are members of the set {00, 01, . . . , 99}. Compute the number of
ordered pairs (A, B) such that |r s| = k 2 for some integer k.
Solution 7.

Because |(10A+B) (10B+A)| = 9|A B| = k 2 , it follows that |A B| is a perfect square.

|A B| = 0
|A B| = 1
|A B| = 4
|A B| = 9

yields
yields
yields
yields

10 pairs of integers: (A, B) = (0, 0), (1, 1), . . . , (9, 9).

18 pairs: the nine (A, B) = (0, 1), (1, 2), . . . , (8, 9), and their reverses.
12 pairs: the six (A, B) = (0, 4), (1, 5), . . . , (5, 9), and their reverses.
2 pairs: (A, B) = (0, 9) and its reverse.

Thus the total number of possible ordered pairs (A, B) is 10 + 18 + 12 + 2 = 42.

Problem 8. For k 3, we define an ordered k-tuple of real numbers (x1 , x2 , . . . , xk ) to be special if, for every i
such that 1 i k, the product x1 x2 . . . xk = x2i .Compute the smallest value of k such that there are at
least 2009 distinct special k-tuples.
Solution 8. The given conditions imply k equations. By taking the product of these k equations, we have
(x1 x2 . . . xk )k1 = x1 x2 . . . xk . Thus it follows that either x1 x2 . . . xk = 0 or x1 x2 . . . xk = 1. If x1 x2 . . . xk = 0,
then some xj = 0, and by plugging this into each of the equations, it follows that all of the xi s are equal to 0.
Note that we cannot have x1 x2 . . . xk = 1, because the left hand side equals x1 (x2 . . . xk ) = x21 , which cant

be negative, because the xi s are all given as real. Thus x1 x2 . . . xk = 1, and it follows that each xi is equal to
either 1 or 1. Because the product of the xi s is 1, there must be an even number of 1s. Furthermore, by
picking any even number of the xi s to be 1, it can be readily verified that the ordered k-tuple (x1 , x2 , . . . , xk )
is special. Thus there are
     


k
k
k
k
+
+
+ ... +
0
2
4
2bk/2c
special non-zero k-tuples. By considering the binomial expansion of (1 + 1)k + (1 1)k , it is clear that the
above sum of binomial coefficients equals 2k1 . Thus there are a total of 2k1 + 1 special k-tuples. Because
210 = 1024 and 211 = 2048, the inequality 2k1 + 1 2009 is first satisfied when k = 12.

Alternate Solution: Use a recursive approach. Let Sk denote the number of special non-zero k-tuples.
From the analysis in the above solution, each xi must be either 1 or 1. It can easily be verified that S3 = 4.
For k > 3, suppose that xk = 1 for a given special k-tuple. Then the k equations that follow are precisely the
equation x1 x2 . . . xk1 = 1 and the k 1 equations that follow for the special (k 1)-tuple (x1 , x2 , . . . , xk1 ).
Because x1 x2 . . . xk1 = 1 is consistent for a special (k 1)-tuple, and because this equation imposes no further
restrictions, we conclude that there are Sk1 special k-tuples in which xk = 1.
If, on the other hand, xk = 1 for a given special k-tuple, then consider the k equations that result, and make
the substitution x1 = y1 . Then the k resulting equations are precisely the same as the k equations obtained
in the case where xk = 1, except that x1 is replaced by y1 . Thus (x1 , x2 , . . . , xk1 , 1) is special if and only if
(y1 , x2 , . . . , xk1 ) is special, and thus there are Sk1 special k-tuples in which xk = 1.
Thus the recursion becomes Sk = 2Sk1 , and because S3 = 4, it follows that Sk = 2k1 .
Problem 9.

A cylinder with radius r and height h has volume 1 and total surface area 12. Compute

Solution 9.

Since r2 h = 1, we have h =

1
r

+ h1 .

1
r 2

and r2 = h1 . Consequently,


 
1
2
2
1
1
1
2rh + 2r2 = 12 (2r)
= 12 + = 12 + = 6.
+
2
r2
h
r
h
r
h

Alternate Solution: The total surface area is 2rh + 2r2 = 12 and the volume is r2 h = 1. Dividing, we
2rh+2r 2
obtain 12
= 2r + h2 , thus 1r + h1 = 12
1 =
r 2 h
2 = 6.
Problem 10.

If 6 tan1 x + 4 tan1 (3x) = , compute x2 .

Solution 10. Let z = 1 + xi and w = 1 + 3xi, where i = 1. Then tan1 x = arg z and tan1 (3x) = arg w,
where arg z gives the measure of the angle in standard position whose terminal side passes through z. By
DeMoivres theorem, 6 tan1 x = arg(z 6 ) and 4 tan1 (3x) = arg(w6 ). Therefore the equation 6 tan1 x +
4 tan1 (3x) = is equivalent to z 6 w4 = a, where a is a real number (and, in fact, a < 0). To simplify
somewhat, we can take the square root of both sides, and get z 3 w2 = 0 + bi, where b is a real number. Then
(1 + xi)3 (1 + 3xi)2 = 0 + bi. Expanding each binomial and collecting real and imaginary terms in each factor
yields ((1 3x2 ) + (3x x3 )i)((1 9x2 ) + 6xi) = 0 + bi. In order that the real part of the product be 0, we have

3
.
(1 3x2 )(1 9x2 ) (3x x3 )(6x) = 0. This equation simplifies to 1 30x2 + 33x4 = 0, yielding x2 = 158
33

158 3

1
Notice that 33 1, which would mean that x 1, and so tan (x) 4 , which is too large, since 6 4 > .
(It can be verified that
this value for x yields a value of 3 for the left side of the equation.) Therefore we are

3
left with x2 = 158
.
33

7
To verify that this answer is reasonable, consider that 3 1.73, so that 15 8 3 1.16, and so x2 200
=

1
1
0.035. Then x itself is a little less than 0.2, and so tan x 15 . Similarly, 3x is about 0.6, so tan (3x) is

about 6 . 6 15
+ 4 6 is reasonably close to .

Alternate Solution: Recall that tan(a + b) =

tan(3a) = tan(2a + a) =

2 tan a
1tan2 a +
2 tan a
1tan
2a

tan a+tan b
1tan a tan b ,

tan a
tan a

thus tan(2a) =

2 tan a
1tan2 a

and

3 tan a tan3 a
2 tan a + tan a tan3 a
=
.
2
2
1 tan a 2 tan a
1 3 tan2 a

Back to the problem at hand, divide both sides by 2 to obtain 3 tan1 x + 2 tan1 (3x) =
tangent of the left side yields
tan

tan(3 tan
x)+tan(2 tan (3x))
1tan(3 tan1 x) tan(2 tan1 (3x)) .
1
1

is undefined, thus 1 = tan(3 tan

x) tan(2 tan

2.

Taking the

We know that the denominator must be 0 since

(3x)) =

3xx3
13x2

23x
1(3x)2

and hence (1 3x2 )(1 9x2 ) =

3
.
(3x x3 )(6x). Simplifying yields 33x4 30x2 + 1 = 0, and applying the quadratic formula gives x2 = 158
33

15+8 3
The + solution is extraneous: as noted in the previous solution, x = 33 yields a value of 3 for the left

3
side of the equation), so we are left with x2 = 158
.
33

Team Problems

Problem 1. Let N be a six-digit number formed by an arrangement of the digits 1, 2, 3, 3, 4, 5. Compute the
smallest value of N that is divisible by 264.
Problem 2. In triangle ABC, AB = 4, BC = 6, and AC = 8. Squares ABQR and BCST are drawn external to
and lie in the same plane as 4ABC. Compute QT .
Problem 3. The numbers 1, 2, . . . , 8 are placed in the 3 3 grid below, leaving exactly one blank square. Such
a placement is called okay if in every pair of adjacent squares, either one square is blank or the difference
between the two numbers is at most 2 (two squares are considered adjacent if they share a common side). If reflections, rotations, etc. of placements are considered distinct, compute the number of distinct okay placements.

Problem 4. An ellipse in the first quadrant is tangent to both the x-axis and y-axis. One focus is at (3, 7), and
the other focus is at (d, 7). Compute d.
Problem 5. Let A1 A2 A3 A4 A5 A6 A7 A8 be a regular octagon. Let u be the vector from A1 to A2 and let v be the
vector from A1 to A8 . The vector from A1 to A4 can be written as au + bv for a unique ordered pair of real
numbers (a, b). Compute (a, b).
Problem 6.
1024.

Compute the integer n such that 2009 < n < 3009 and the sum of the odd positive divisors of n is

Problem 7. Points A, R, M , and L are consecutively the midpoints of the sides of a square whose area is 650.
The coordinates of point A are (11, 5). If points R, M , and L are all lattice points, and R is in Quadrant I,
compute the number of possible ordered pairs (x, y) of coordinates for point R.

Problem 8.

The taxicab distance between points (x1 , y1 , z1 ) and (x2 , y2 , z2 ) is given by

d((x1 , y1 , z1 ), (x2 , y2 , z2 )) = |x1 x2 | + |y1 y2 | + |z1 z2 |.

The region R is obtained by taking the cube {(x, y, z) : 0 x, y, z 1} and removing every point whose
taxicab distance to any vertex of the cube is less than 53 . Compute the volume of R.
Problem 9.

Let a and b be real numbers such that

a3 15a2 + 20a 50 = 0

and 8b3 60b2 290b + 2575 = 0.

Compute a + b.
Problem 10. For a positive integer n, define s(n) to be the sum of n and its digits. For example, s(2009) =
2009 + 2 + 0 + 0 + 9 = 2020. Compute the number of elements in the set {s(0), s(1), s(2), . . . , s(9999)}.

Team Answers

Answer 1. 135432

Answer 2. 2 10
Answer 3. 32
Answer 4.

49
3

Answer 5. 2 +

2, 1 +

2

Answer 6. 2604
Answer 7. 10
Answer 8.

179
250

Answer 9.

15
2

Answer 10. 9046

Team Solutions

Problem 1. Let N be a six-digit number formed by an arrangement of the digits 1, 2, 3, 3, 4, 5. Compute the
smallest value of N that is divisible by 264.
Solution 1. 264 = 3 8 11, so we will need to address all these factors. Because the sum of the digits is 18, it
follows that 3 divides N , regardless of how we order the digits of N . In order for 8 to divide N , we need N
to end in O12, O52, E32, or E24, where O and E denote odd and even digits. Now write N = U V W X Y Z.
Note that N is divisible by 11 if and only if (U + W + Y ) (V + X + Z) is divisible by 11. Because the sum
of the three largest digits is only 12, we must have U + W + Y = V + X + Z = 9.
Because Z must be even, this implies that V, X, Z are 2, 3, 4 (in some order). This means Y 6= 2, and so we
must have Z 6= 4 Z = 2. Of the three remaining possibilities, E32 gives the smallest solution, 135432.

Problem 2. In triangle ABC, AB = 4, BC = 6, and AC = 8. Squares ABQR and BCST are drawn external to
and lie in the same plane as 4ABC. Compute QT .

T
Q
S
R

B
C

Solution 2. Set mABC = x and mT BQ = y. Then x + y = 180 and so cos x + cos y = 0. Applying the Law
of Cosines to triangles ABC and T BQ gives AC 2 = AB 2 + BC 2 2AB BC cos x and QT 2 = BT 2 + BQ2
2BT BQ cos y, which, after substituting values, become 82 = 42 + 62 48 cos x and QT 2 = 42 + 62 48 cos y.

Adding the last two equations yields QT 2 + 82 = 2(42 + 62 ) or QT = 2 10.

Remark: This problem is closely related to the fact that in a parallelogram, the sum of the squares of the
lengths of its diagonals is the equal to the sum of the squares of the lengths of its sides.
Problem 3. The numbers 1, 2, . . . , 8 are placed in the 3 3 grid below, leaving exactly one blank square. Such
a placement is called okay if in every pair of adjacent squares, either one square is blank or the difference
between the two numbers is at most 2 (two squares are considered adjacent if they share a common side). If reflections, rotations, etc. of placements are considered distinct, compute the number of distinct okay placements.

Solution 3. We say that two numbers are neighbors if they occupy adjacent squares, and that a is a friend of b if
0 < |a b| 2. Using this vocabulary, the problems condition is that every pair of neighbors must be friends
of each other. Each of the numbers 1 and 8 has two friends, and each number has at most four friends.
If there is no number written in the center square, then we must have one of the cycles in the figures below.
For each cycle, there are 8 rotations. Thus there are 16 possible configurations with no number written in the
center square.

Now assume that the center square contains the number n. Because n has at least three neighbors, n 6= 1 and
n 6= 8. First we show that 1 must be in a corner. If 1 is a neighbor of n, then one of the corners neighboring
1 must be empty, because 1 has only two friends (2 and 3). If c is in the other corner neighboring 1, then
{n, c} = {2, 3}. But then n must have three more friends (n1 , n2 , n3 ) other than 1 and c, for a total of five
friends, which is impossible, as illustrated below. Therefore 1 must be in a corner.

n1

n2

n3
Now we show that 1 can only have one neighbor, i.e., one of the squares adjacent to 1 is empty. If 1 has two
neighbors, then we have, up to a reflection and a rotation, the configuration shown below. Because 2 has only
one more friend, the corner next to 2 is empty and n = 4. Consequently, m1 = 5 (refer to the figure below).
Then 4 has one friend (namely 6) left with two neighbors m2 and m3 , which is impossible. Thus 1 must have
exactly one neighbor. An analogous argument shows that 8 must also be at a corner with exactly one neighbor.

m3

m1 m2
Therefore, 8 and 1 must be in non-opposite corners, with the blank square between them. Thus, up to reflections
and rotations, the only possible configuration is the one shown at left below.

2/3 4/5 6/7

3/2 5/4 7/6

There are two possible values for m, namely 2 and 3. For each of the cases m = 2 and m = 3, the rest of the
configuration is uniquely determined, as illustrated in the figure above right. We summarize our process: there
are four corner positions for 1; two (non-opposite) corner positions for 8 (after 1 is placed); and two choices for

10

the number in the square neighboring 1 but not neighboring 8. This leads to 4 2 2 = 16 distinct configurations
with a number written in the center square.
Therefore, there are 16 configurations in which the center square is blank and 16 configurations with a number
in the center square, for a total of 32 distinct configurations.
Problem 4. An ellipse in the first quadrant is tangent to both the x-axis and y-axis. One focus is at (3, 7), and
the other focus is at (d, 7). Compute d.


Solution 4. See the diagram below. The center of the ellipse is C = d+3
2 , 7 . The major axis of the ellipse
is the line y = 7, and the minor axis is the line x = d+3
2 . The ellipse is tangent to the coordinate axes at


d+3
Tx = 2 , 0 and Ty = (0, 7). Let F1 = (3, 7) and F2 = (d, 7). Using the locus definition of an ellipse, we have
F1 Tx + F2 Tx = F1 Ty + F2 Ty ; that is,
s
2
p
d3
+ 72 = d + 3 or
2
(d 3)2 + 142 = d + 3.
2
Squaring both sides of the last equation gives d2 6d + 205 = d2 + 6d + 9 or 196 = 12d, so d =

Ty

F1

49
3 .

F2

Tx

Problem 5. Let A1 A2 A3 A4 A5 A6 A7 A8 be a regular octagon. Let u be the vector from A1 to A2 and let v be the
vector from A1 to A8 . The vector from A1 to A4 can be written as au + bv for a unique ordered pair of real
numbers (a, b). Compute (a, b).

Solution 5. We can scale the octagon so that A1 A2 = 2. Because the exterior angle
of the octagon is 45 , we

can place the octagon in the coordinate plane with A1 being the origin, A2 = ( 2, 0), and A8 = (1, 1).

A6

A5

A7

A4

A8

A3
A1

Then A3 = (1 +

A2


2, 1) and A4 = (1 + 2, 1 + 2). It follows that u =
2, 0 , v = h1, 1i, and
D
E
D
E
D
E

A1 A4 = 1 + 2, 1 + 2 = a
2, 0 + b h1, 1i = a 2 b, b .

11

Thus b =

2 + 1 and a 2 b = 2 + 1, or a = 2 + 2.

Alternate Solution: Extend A1 A2 and A5 A4 to meet at point Q; let P be the intersection of A1 Q and

A6 A3 . Then A1 A2 = kuk, A2 P = kuk 2, and P Q = kuk, so A1 Q = 2 + 2 kuk. Because A1 QA4 is a

45 45 90 right triangle, A4 Q = A12Q =
2 + 1 kuk. Thus A1 A4 = A1 Q + QA4 , and because kuk = kvk,

we have (a, b) = (2 + 2, 2 + 1).
Problem 6. Compute the integer n such that 2009 < n < 3009 and that the sum of the odd positive divisors of
n is 1024.
Solution 6. Suppose that n = 2k pa1 1 par r , where the pi are distinct odd primes, k is a nonnegative integer, and
a1 , . . . , ar are positive integers. Then the sum of the odd positive divisors of n is equal to
r
Y

(1 + pi + + pai i ) =

i=1

r
Y
pai +1 1
i

i=1

pi 1

= 1024 = 210 .

Note that 1 + pi + + pai i is the sum of ai + 1 odd numbers. Because the product of those sums is a power
of two, each sum must be even (in fact, a power of 2). Thus, each ai must be odd.
Because 1 + 11 + 112 + 113 > 1024, if pi 11, then ai = 1 and 1 + pi must be a power of 2 that is no greater
than 1024. The possible values of pi , with pi 11, are 31 and 127 (as 5 divides 255, 7 divides 511, and 3
divides 1023).
If p1 < 11, then pi can be 3, 5, 7. It is routine to check that ai = 1 and pi = 3 or 7.
Thus ai = 1 for all i, and the possible values of pi are 3, 7, 31, 127. The only combinations of these primes
that yield 1024 are (1 + 3) (1 + 7) (1 + 31) (with n = 2k 3 7 31 = 651 2k ) and (1 + 7) (1 + 127) (with
n = 7 127 = 889 2k ). Thus n = 651 22 = 2604 is the unique value of n satisfying the conditions of the
problem.
Problem 7. Points A, R, M , and L are consecutively the midpoints of the sides of a square whose area is 650.
The coordinates of point A are (11, 5). If points R, M , and L are all lattice points, and R is in Quadrant I,
compute all possible ordered pairs (x, y) of coordinates for point R.
Solution 7. Write x = 11 + c and y = 5 + d. Then AR2 = c2 + d2 = 21 650 = 325. Note that 325 = 182 + 12 =
172 + 62 = 152 + 102 . Temporarily restricting ourselves to the case where c and d are both positive, there
are three classes of solutions: {c, d} = {18, 1}, {c, d} = {17, 6}, or {c, d} = {15, 10}. In fact, c and d can be
negative, so long as those values do not cause x or y to be negative. So there are ten solutions:
(c, d)
(18, 1)
(18, 1)
(1, 18)
(1, 18)
(17, 6)
(6, 17)
(6, 17)
(15, 10
(10, 15)
(10, 15)

12

(x, y)
(29, 6)
(29, 4)
(12, 23)
(10, 23)
(28, 11)
(17, 22)
(5, 22)
(26, 15)
(21, 20)
(1, 20)

Problem 8.

The taxicab distance between points (x1 , y1 , z1 ) and (x2 , y2 , z2 ) is given by

d((x1 , y1 , z1 ), (x2 , y2 , z2 )) = |x1 x2 | + |y1 y2 | + |z1 z2 |.

The region R is obtained by taking the cube {(x, y, z) : 0 x, y, z 1} and removing every point whose
taxicab distance to any vertex of the cube is less than 53 . Compute the volume of R.
Solution 8. For a fixed vertex V on the cube, the locus of points on or inside the cube that are at most 35 away from
V form a corner at V (that is, the right pyramid V W1 W2 W3 in the figure shown at left below, with equilateral
triangular base W1 W2 W3 and three isosceles right triangular lateral faces V W1 W2 , V W2 W3 , V W3 W1 ). Thus
R is formed by removing eight such congruent corners from the cube. However, each two neighboring corners
share a common region along their shared edge. This common region is the union of two smaller right pyramids,
each similar to the original corners. (See the figure shown at right below.)

W2
W3
V

W1

We compute the volume of R as

1
18
6

Problem 9.

 3
 3
3
1 1
179
+ 12 2
=
.
5
6 10
250

Let a and b be real numbers such that

a3 15a2 + 20a 50 = 0

and 8b3 60b2 290b + 2575 = 0.

Compute a + b.
Solution 9. Each cubic expression can be depressedthat is, the quadratic term can be eliminatedby substituting as follows. Because (a p)3 = a3 3a2 p + 3ap2 p3 , setting p = (15)
= 5 and substituting c + p = a
3
transforms the expression a3 15a2 +20a50 into the equivalent expression (c+5)3 15(c+5)2 +20(c+5)50,
which simplifies to c3 55c 200. Similarly, the substitution d = b 52 yields the equation d3 55d = 200.
[This procedure, which is analogous to completing the square, is an essential step in the algebraic solution to
the general cubic equation.]

Consider the function f (x) = x3 55x. It has three zeros,

namely, 0 and 55. Therefore, it has a relative
maximum and a relative minimumin the interval [ 55, 55]. Note that for 0 x 5.5, |f (x)| < |x3 | <
5.53 = 166.375, and for 5.5 < x 55 < 8, we have
|f (x)| = |x3 55x| < x|x2 55| < 8(55 5.52 ) = 198.

13


Because
f (x) is an odd function of x (its graph is symmetric about the origin), we conclude that for 55
x 55, |f (x)| < 198. Therefore, for constant m with |m| > 198, there is a unique real number x0 such that
f (x0 ) = m.
In particular, since 200 > 198, the values of c and d are uniquely determined. Because f (x) is odd, we conclude
that c = d, or a + b = 15
2 .
Alternate Solution: Set a = x b and substitute into the first equation. We get
(x b)3 15(x b)2 + 20(x b) 50 = 0
b3 + b2 (3x 15) + b(3x2 + 30x 20) + (x3 15x2 + 20x 50) = 0
8b3 + b2 (24x + 120) + b(24x2 240x + 160) 8(x3 15x2 + 20x 50) = 0.
If we equate coefficients, we see that
24x + 120 = 60
2

24x 240x + 160 = 290

3

8(x 15x2 + 20x 50) = 2575

are all satisfied by x = 15
2 . This means that any real solution b to the second equation yields a real solution of
15

b
to
the
first
equation.
We can follow the reasoning of the previous solution to establish the existence of
2
15
exactly one real solution to the second cubic equation. Thus a and b are unique, and their sum is ( 15
2 b)+b = 2 .
Problem 10. For a positive integer n, define s(n) to be the sum of n and its digits. For example, s(2009) =
2009 + 2 + 0 + 0 + 9 = 2020. Compute the number of elements in the set {s(0), s(1), s(2), . . . , s(9999)}.
Solution 10. If s(10x) = a, then the values of s over {10x + 0, 10x + 1, . . . , 10x + 9} are a, a + 2, a + 4, . . . , a + 18.
Furthermore, if x is not a multiple of 10, then s(10(x + 1)) = a + 11. This indicates that the values of s
interweave somewhat from one group of 10 to the next: the sets alternate between even and odd. Because
the s-values for starting blocks of ten differ by 11, consecutive blocks of the same parity differ by 22, so the
values of s do not overlap. That is, s takes on 100 distinct values over any range of the form {100y + 0, 100y +
1, . . . , 100y + 99}.
First determine how many values are repeated between consecutive hundreds. Let y be an integer that is not
a multiple of 10. Then the largest value for s(100y + k) (0 k 99) is 100y + (s(y) y) + 99 + s(99) =
100y + s(y) y + 117, whereas the smallest value in the next group of 100 is for
s(100(y + 1))

100(y + 1) + (s(y + 1) (y + 1)) = 100y + (s(y) + 2) (y + 1) + 100

100y + s(y) y + 101.

This result implies that the values for s(100y + 91) through s(100y + 99) match the values of s(100y + 100)
through s(100y + 108). So there are 9 repeated values.
Now determine how many values are repeated between consecutive thousands. Let z be a digit, and consider
s(1000z + 999) versus s(1000(z + 1)). The first value equals
1000z + (s(z) z) + 999 + s(999) = 1000z + z + 1026 = 1001z + 1026.
The latter value equals 1000(z + 1) + (s(z + 1) (z + 1)) = 1001(z + 1) = 1001z + 1001. These values differ
by an odd number. We have overlap between the 982, 983, . . . , 989 terms and the 000, 001, . . . , 007 terms. We
also have overlap between the 992, 993, . . . , 999 terms and the 010, 011, . . . , 017 terms, for a total of 16 repeated
values in all.
There are 90 instances in which we have 9 repeated terms, and 9 instances in which we have 16 repeated terms,
so there are a total of 10000 90 9 9 16 = 9046 unique values.

14

Power Question: Sign on the Label

Instructions: The power question is worth 50 points; each parts point value is given in brackets next to the
part. To receive full credit the presentation must be legible, orderly, clear, and concise. If a problem says list or
compute, you need not justify your answer. If a problem says determine or find, then you must show your
work or explain your reasoning to receive full credit, although such explanations do not have to be lengthy. If a
problem says justify, then you must prove your answer rigorously. Even if not proved, earlier numbered items may
be used in solutions to later numbered items but not vice-versa. Pages submitted for credit should be NUMBERED
IN CONSECUTIVE ORDER AT THE TOP OF EACH PAGE in what your team considers to be proper sequential
order. PLEASE WRITE ON ONLY ONE SIDE OF THE ANSWER PAPERS. Put the TEAM NUMBER (not the
Team name) on the cover sheet used as the first page of the papers submitted. Do not identify the Team in any
other way.
An n-label is a permutation of the numbers 1 through n. For example, J = 35214 is a 5-label and K = 132 is
a 3-label. For a fixed positive integer p, where p n, consider consecutive blocks of p numbers in an n-label. For
example, when p = 3 and L = 263415, the blocks are 263, 634, 341, and 415. We can associate to each of these blocks
a p-label that corresponds to the relative order of the numbers in that block. For L = 263415, we get the following:
263415 132;

263415 312;

263415 231;

263415 213.

Moving from left to right in the n-label, there are np+1 such blocks, which means we obtain an (np+1)-tuple
of p-labels. For L = 263415, we get the 4-tuple (132, 312, 231, 213). We will call this (np+1)-tuple the p-signature
of L (or signature, if p is clear from the context) and denote it by Sp [L]; the p-labels in the signature are called
windows. For L = 263415, the windows are 132, 312, 231, and 213, and we write
S3 [263415] = (132, 312, 231, 213).
More generally, we will call any (n p + 1)-tuple of p-labels a p-signature, even if we do not know of an n-label
to which it corresponds (and even if no such label exists). A signature that occurs for exactly one n-label is called
unique, and a signature that doesnt occur for any n-labels is called impossible. A possible signature is one that
occurs for at least one n-label.
In this power question, you will be asked to analyze some of the properties of labels and signatures.

The Problems
1. (a) Compute the 3-signature for 52341.
(b) Find another 5-label with the same 3-signature as in part (a).
(c) Compute two other 6-labels with the same 4-signature as 462135.

[1]
[2]
[2]

2. (a)
(b)
(c)
(d)

[1]
[1]
[1]
[1]

Explain why the label 1234 has a unique 3-signature.

List three other 4-labels with unique 3-signatures.
Explain why the 3-signature (123, 321) is impossible.
List three other impossible 3-signatures that have exactly two windows.

We can associate a shape to a given 2-signature: a diagram of up and down steps that indicates the relative order of adjacent numbers. For example, the following shape corresponds to the 2-signature (12, 12, 12, 21, 12, 21):

A 7-label with this 2-signature corresponds to placing the numbers 1 through 7 at the nodes above so that
numbers increase with each up step and decrease with each down step. The 7-label 2347165 is shown below:
6

7
4

3
2

3. Consider the shape below:

(a) Find the 2-signature that corresponds to this shape.

[2]

(b) Compute two different 6-labels with the 2-signature you found in part (a).

[3]

4. (a) List all 5-labels with 2-signature (12, 12, 21, 21).

[2]

(b) Find a formula for the number of (2n + 1)-labels with the 2-signature

[3]

(12, 12, . . . , 12, 21, 21, . . . , 21).

{z
} |
{z
}
|
n

5. (a) Compute the number of 5-labels with 2-signature (12, 21, 12, 21).

[2]

(b) Determine the number of 9-labels with 2-signature

(12, 21, 12, 21, 12, 21, 12, 21).
Justify your answer.

[3]

6. (a) Determine whether the following signatures are possible or impossible:

(i) (123, 132, 213),
(ii) (321, 312, 213).

[1]
[1]

(b) Notice that a (p + 1)-label has only two windows in its p-signature. For a given window 1 , compute the
number of windows 2 such that Sp [L] = (1 , 2 ) for some (p + 1)-label L.
[1]
(c) Justify your answer from part (b).

[2]

7. (a) For a general n, determine the number of distinct possible p-signatures.

[4]

(b) If a randomly chosen p-signature is 575 times more likely of being impossible than possible, determine p
and n.
[1]
8. (a) Show that (312, 231, 312, 132) is not a unique 3-signature.

[1]

(b) Show that (231, 213, 123, 132) is a unique 3-signature.

[2]

(c) Find two 5-labels with unique 2-signatures.

[1]

16

(d) Find a 6-label with a unique 4-signature but which has the 3-signature from part (a).

[2]

9. (a) For a general n 2, compute all n-labels that have unique 2-signatures.

[1]

(b) Determine whether or not S5 [495138627] is unique.

[1]

(c) Determine the smallest p for which the 20-label

L = 3, 11, 8, 4, 17, 7, 15, 19, 6, 2, 14, 1, 10, 16, 5, 12, 20, 9, 13, 18
has a unique p-signature.

[3]
k

10. Show that for each k 2, the number of unique 2k1 -signatures on the set of 2k -labels is at least 22

17

[5]

Power Solutions
1 (a) (312, 123, 231)
(b) There are three: 41352, 42351, 51342.
(c) There are three: 362145, 452136, 352146.
2 (a) We will prove this by contradiction. Suppose for some other 4-label L we have S3 [L] = S3 [1234] =
(123, 123). Write out L as a1 , a2 , a3 , a4 . From the first window of S3 [L], we have a1 < a2 < a3 . From the
second window, we have a2 < a3 < a4 . Connecting these inequalities gives a1 < a2 < a3 < a4 , which
forces L = 1234, a contradiction. Therefore, the 3-signature above is unique.
(b) There are 11 others (12 in all, if we include S3 [1234]):
S3 [1234] = (123, 123)

S3 [1243] = (123, 132)

S3 [1324] = (132, 213)

S3 [1423] = (132, 213)

S3 [2134] = (213, 123)

S3 [2314] = (231, 213)

S3 [3241] = (213, 231)

S3 [3421] = (231, 321)

S3 [4132] = (312, 132)

S3 [4231] = (312, 231)

S3 [4312] = (321, 312)

S3 [4321] = (321, 321)

(c) If S3 [a1 , a2 , a3 , a4 ] = (123, 321), then the first window forces a2 < a3 , whereas the second window forces
a2 > a3 . This is impossible, so the 3-signature (123, 321) is impossible.
(d) There are 18 impossible 3-signatures with two windows. In nine of these, the first window indicates that
a2 < a3 (an increase), but the second window indicates that a2 > a3 (a decrease). In the other nine, the
end of the first window indicates a decrease (that is, a2 > a3 ), but the beginning of the second window
indicates an increase (a2 < a3 ). In general, for a 3-signature to be possible, the end of the first window
and beginning of the second window must be consistent, indicating either an increase or a decrease. The
impossible 3-signatures are
(123, 321)

(123, 312)

(123, 213)

(132, 231)

(132, 132)

(132, 123)

(213, 321)

(213, 312)

(213, 213)

(231, 231)

(231, 132)

(231, 123)

(312, 321)

(312, 312)

(312, 213)

(321, 231)

(321, 132)

(321, 123)

3. (a) The first pair indicates an increase; the next three are decreases, and the last pair is an increase. So the
2-signature is (12, 21, 21, 21, 12).
(b) There are several:

564312

564213

563214

465312

465213

463215

365412

365214

364215

265413

265314

264315

165413

165314

164315

354216

254316

154326

453216

4 In part (a), we can count by brute force, or use the formula from part (b) (with indepedent proof).
(a) For the case of 5-labels, brute force counting is tractable.
12543, 13542, 14532, 23541, 24531, 34521 .
 
2n
(b) The answer is
.
n
The shape of this signature is a wedge: n up steps followed by n down steps. The wedge for n = 3 is
illustrated below:

18

The largest number in the label, 2n + 1, must be placed at the peak in the center. If we choose the numbers to put in the first n spaces, then they must be placed in increasing order. Likewise, the remaining n
numbers must

be placed in decreasing order on the downward sloping piece of the shape. Thus there are
exactly 2n
n such labels.
5 (a) The answer is 16 .
We have a shape with two peaks and a valley in the middle. The 5 must go on one of the two peaks, so
we place it on the first peak. By the shapes symmetry, we will double our answer at the end to account
for the 5-labels where the 5 is on the other peak.

The 4 can go to the left of the 5 or at the other peak. In the first case, shown below left, the 3 must go
at the other peak and the 1 and 2 can go in either order. In the latter case, shown below right, the 1, 2,
and 3 can go in any of 3! arrangements.

So there are 2! + 3! = 8 possibilities. In all, there are 16 5-labels (including the ones where the 5 is at the
other peak).
(b) The answer is 7936 .
The shape of this 2-signature has four peaks and three intermediate valleys:

We will solve this problem by building up from smaller examples. Let fn equal the number of (2n + 1)labels whose 2-signature consists of n peaks and n 1 intermediate valleys. In part (b) we showed that
f2 = 16. In the case where we have one peak, f1 = 2. For the trivial case (no peaks), we get f0 = 1.
These cases are shown below.

19

1
1

Suppose we know the peak on which the largest number, 2n + 1, is placed. Then that splits our picture
into two shapes with fewer peaks. Once we choose which numbers from 1, 2, . . . , 2n to place each shape,
we can compute the number of arrangments of the numbers on each shape, and then take the product.
For example, if we place the 9 at the second peak, as shown below, we get a 1-peak shape on the left and
a 2-peak shape on the right.

9


For the above shape, there are 83 ways to pick the three numbers to place on the lefthand side, f1 = 2
ways to place them, and f2 = 16 ways to place the remaining five numbers on the right.
This argument works for any n > 1, so we have shown the following:

n 
X
2n
fn =
fk1 fnk .
2k 1
k=1

So we have:
f1
f2
f3
f4

 
2 2
=
f =2
1 0
 
 
4
4
=
f0 f1 +
f1 f0 = 16
1
3
 
 
 
6
6 2
6
=
f0 f2 +
f +
f2 f0 = 272
1
3 1
5
 
 
 
 
8
8
8
8
=
f0 f3 +
f1 f2 +
f2 f1 +
f3 f0 = 7936 .
1
3
5
7

6 (a) Signature (i) is possible, because it is the 3-signature of 12435.

Signature (ii) is impossible. Let a 5-label be a1 , a2 , a3 , a4 , a5 . The second window of (ii) implies a3 < a4 ,
whereas the third window implies a3 > a4 , a contradiction.
(b) There are p such windows, regardless of 1 .
(c) Because the windows 1 and 2 overlap for p 1 numbers in L, the first p 1 integers in 2 must be
in the same relative order as the last p 1 integers in 1 . Therefore, by choosing the last integer in 2
(there are p choices), the placement of the remaining p 1 integers is determined. We only need to show
that there exists some L such that Sp [L] = (1 , 2 ).
To do so, set 1 = w1 , w2 , . . . , wp . Provisionally, let L(k) = w1 , w2 , . . . , wp , k + 0.5 for k = 0, 1, . . . , p. For
example, if p = 4 and 1 = 3124, then L(2) = 3, 1, 2, 4, 2.5. First we show that L(k) has the required

20

p-signature; then we can renumber the entries in L(k) in consecutive order to make them all integers, so
for example 3, 1, 2, 4, 2.5 would become 4, 1, 2, 5, 3.
Even though the last entry in L(k) is not an integer, we can still compute Sp [L(k) ], since all we require
is that the entries are all distinct. In each Sp [L(k) ], the first window is 1 . When we compare L(k) to
L(k+1) , the last integer in 2 can increase by at most 1, because the last integer in the label jumps over
at most one integer in positions 2 through p (those jumps are in boldface):
L(0) = 3, 1, 2, 4, 0.5

S4 [L(0) ] = (3124, 2341)

L(1) = 3, 1, 2, 4, 1.5

S4 [L(1) ] = (3124, 1342)

(2)

= 3, 1, 2, 4, 2.5

L(3) = 3, 1, 2, 4, 3.5
(4)

= 3, 1, 2, 4, 4.5

(2)

S4 [L

] = (3124, 1243)

S4 [L(3) ] = (3124, 1243)

(4)

S4 [L

] = (3124, 1234)

[jumps over the 1]

[jumps over the 2]
[jumps over nothing]
[jumps over the 4]

The final entry of 2 in Sp [L(0) ] is 1, because 0.5 is smaller than all other entries in L(0) . Likewise, the
final entry of 2 in Sp [L(p) ] is p. Since the final entry increases in increments of 0 or 1 (as underlined
above), we must see all p possibilities for 2 .
By replacing the numbers in each L(k) with the integers 1 through p + 1 (in the same relative order as the
numbers in L(k) ), we have found the (p + 1)-labels that yield all p possibilities.
7 (a) The answer is p! pnp .
Call two consecutive windows in a p-signature compatible if the last p 1 numbers in the first label and
the first p 1 numbers in the second label (their overlap) describe the same ordering. For example, in
the p-signature (. . . , 2143, 2431, . . .), 2143 and 2431 are compatible. Notice that the last three digits of
2143 and the first three digits of 2431 can be described by the same 3-label, 132.
Theorem. A signature is possible if and only if every pair of consecutive windows is compatible.
Proof. () Consider a signature describing a p-label L. If some pair in is not compatible, then there
is some string of p 1 numbers in our label L that has two different (p 1)-signatures. This is impossible,
since the p-signature is well-defined.
() Now suppose is a p-signature such that that every pair of consecutive windows is compatible. We
need to show that there is at least one label L with Sp [L] = . We do so by induction on the number of
windows in , using the results from 5(b).
Let = {1 , 2 , . . . , k+1 }, and suppose 1 = a1 , a2 , . . . , ap . Set L1 = 1 .
Suppose that Lk is a (p + k 1)-label such that Sp [Lk ] = {1 , . . . , k }. We will construct Lk+1 for which
Sp [Lk+1 ] = {1 , . . . , k+1 }.
(j)

As in 5(b), denote by Lk the label Lk with a j + 0.5 appended; we will eventually renumber the elements in the label to make them all integers. Appending j + 0.5 does not affect any of the non-terminal
(j)
windows of Sp [Lk ], and as j varies from 0 to p k + 1 the final window of Sp [Lk ] varies over each of
the p windows compatible with k . Since k+1 is compatible with k , there exists some j for which
(j)
(j)
(j)
Sp [Lk ] = {1 , . . . , k+1 }. Now we renumber as follows: set Lk+1 = Sk+p [Lk ], which replaces Lk with
the integers 1 through k + p and preserves the relative order of all integers in the label.
By continuing this process, we conclude that the n-label Lnp+1 has p-signature , so is possible.

21

To count the number of possible p-signatures, we choose the first window (p! choices), then choose each
of the remaining np compatible windows (p choices each). In all, there are p!pnp possible p-signatures.
(b) The answer is n = 7, p = 5 .
Let P denote the probability that a randomly chosen p-signature is possible. We are given that 1P = 575,
1
. We want to find p and n for which
so P = 576
p! pnp
1
=
(p!)np+1
576
1
pnp
=
(p!)np
576
((p 1)!)np = 576.
The only factorial that has 576 as an integer power is 4! =

576. Thus p = 5 and n p = 2 n = 7.

8 (a) The p-signature is not unique because it equals both S3 [625143] and S3 [635142].
(b) Let L = a1 , a2 , a3 , a4 , a5 , a6 . We have a4 < a6 < a5 (from window #4), a3 < a1 < a2 (from window #1),
and a2 < a4 (from window #2). Linking these inequalities, we get
a3 < a1 < a2 < a4 < a6 < a5

L = 231456,

so S3 [L] is unique.
(c) 12345 and 54321 are the only ones.
(d) L = 645132 will work. First, note that S3 [645132] = (312, 231, 312, 132). Next, we need to show
that S4 [645132] = {4231, 3412, 4132} is unique. So let L0 = a1 , a2 , a3 , a4 , a5 , a6 be a 6-label such that
S4 [L0 ] = (4231, 3412, 4132).
We get a4 < a6 < a5 (from window #3), a5 < a2 < a3 (from window #2), and a3 < a1 (from window
#1). Linking these inequalities, we get
a4 < a6 < a5 < a2 < a3 < a1

L0 = 645132,

so L0 = L, which means S4 [L] is unique.

9 (a) The n-labels with unique 2-signatures are 1, 2, . . . , n and n, n 1, . . . , 1, and their respective 2-signatures
are (12, 12, . . . , 12) and (21, 21, . . . , 21).
Proof: (Not required for credit.) Let L = a1 , a2 , . . . , an . The first signature above implies that
a1 < a2 < < an , which forces a1 = 1, a2 = 2, and so on. Likewise, the second signature forces
a1 = n, a2 = n 1, and so on.
To show that all other n-labels fail to have unique 2-signatures, we will show that in any other n-label
L0 , there are two numbers k and k + 1 that are not adjacent. By switching k and k + 1 we get a label
L00 for which S2 [L0 ] = S2 [L00 ], since the differences between k and its neighbors in L0 were at least 2 (and
likewise for k + 1).
To show that such a k and k + 1 exist, we proceed by contradiction. Suppose that all such pairs are
adjacent in L0 . Then 1 and n must be at the ends of L0 (or else some intermediate number k will fail to

22

be adjacent to k + 1 or k 1). But if a1 = 1, then that forces a2 = 2, a3 = 3, . . . , an = n. That is, we get

the two labels we already covered. (We get the other label if an = 1.)
Therefore, none of the n! 2 remaining labels has a unique 2-signature.
(b) S5 [495138627] is unique.
Let L = a1 , . . . , a9 and suppose S5 [L] = S5 [495138627] = (1 , . . . , 5 ). Then we get the following inequalities:
a4 < a8

[from 4 ]

a8 < a5

[from 4 ]

a5 < a1

[from 1 ]

a1 < a3

[from 1 ]

a3 < a7

[from 3 ]

a7 < a9

[from 5 ]

a9 < a6

[from 5 ]

a6 < a2

[from 2 ]

Combining, we get a4 < a8 < a5 < a1 < a3 < a7 < a9 < a6 < a2 , which forces a4 = 1, a8 = 2, . . . , a2 = 9.
So the label L is forced and S5 [495138627] is therefore unique.
(c) The answer is p = 16 . To show this fact we will need to extend the idea from part 8(b) about linking
inequalities forced by the various windows:
Theorem: A p-signature for an n-label L is unique if and only if for every k < n, k and k + 1 are in at
least one window together. That is, the distance between them in the n-label is less than p.
Proof. Suppose that for some k, the distance between k and k + 1 is p or greater. Then the label L0
obtained by swapping k and k + 1 has the same p-signature, because there are no numbers between k and
k + 1 in any window and because the two numbers never appear in the same window.
If the distance between all such pairs is less than k, we need to show that Sp [L] is unique. For i = 1, 2, . . . , n,
let ri denote the position where i appears in L. For example, if L = 4123, then r1 = 2, r2 = 3, r3 = 4, and
r4 = 1.
Let L = a1 , a2 , . . . , an . Since 1 and 2 are in some window together, ar1 < ar2 . Similarly, for any k, since k
and k +1 are in some window together, ark < ark+1 . We then get a linked inequality ar1 < ar2 < < arn ,
which can only be satisfied if ar1 = 1, ar2 = 2, . . . , arn = n. Therefore, Sp [L] is unique.
2
From the proof above, we know that the signature is unique if and only if every pair of consecutive integers
coexists in at least one window. Therefore, we seek the largest distance between consecutive integers in
L. That distance is 15 (from 8 to 9, and from 17 to 18). Thus the smallest p is 16 .
10 Let sk denote the number of such unique signatures. We proceed by induction with base case k = 2. From
8(c), a 2-signature for a label L is unique if and only if consecutive numbers in L appear together in some
window. Because k = 2, the consecutive numbers must be adjacent in the label. The 22 -labels 1234 and 4321
2
satisfy this condition, 1 so their 21 -signatures are unique. Thus we have shown that s2 2 = 22 3 , and the
base case is established.
k

Now suppose sk 22 3 for some k 2. Let Lk be a 2k -label with a unique 2k1 -signature. Write
Lk = (a1 , a2 , . . . , a2k ). We will expand Lk to form a 2k+1 label by replacing each ai above with the numbers 2ai 1 and 2ai (in some order). This process produces a valid 2k+1 -label, because the numbers produced
1 The proof in 8(a) establishes that these are the only 22 -labels with unique 2-signatures, thus giving s = 2. That proof was not
2
required for credit and is not needed here, since the inequality above is good enough for this problem.

23

are all the integers from 1 to 2k+1 . Furthermore, different Lk s will produce different labels: if the starting
labels differ at place i, then the new labels will differ at places 2i 1 and 2i. Therefore, each starting label
k
produces 22 distinct 2k+1 -labels through this process. Summarizing, each valid 2k -label can be expanded to
k
produce 22 distinct 2k+1 -labels, none of which could be obtained by expanding any other 2k -label.
It remains to be shown that the new label has a unique 2k1 -signature. Because Lk has a unique 2k1 -signature,
for all i 2k 1, both i and i + 1 appeared in some 2k1 -window. Therefore, there were fewer than 2k1 1
numbers between i and i + 1 . When the label is expanded, 2i and 2i 1 are adjacent, 2i + 1 and 2i + 2 are
adjacent, and 2i and 2i + 1 are fewer than 2 (2k1 1) + 2 = 2k places apart. Thus, every pair of adjacent
integers is within some 2k -window.
Since each pair of consecutive integers in our new 2k+1 -label coexists in some 2k -window for every possible
k
such expansion of Lk , that means all 22 ways of expanding Lk to a 2k+1 -label result in labels with unique
2k -signatures. We then get
k

sk+1 22 sk
k

22 22
= 22
which completes the induction.

24

k+1

Relay Problems

Relay 1-1 A rectangular box has dimensions 8 10 12. Compute the fraction of the boxs volume that is not
within 1 unit of any of the boxs faces.
2

Relay 1-2 Let T = T N Y W R. Compute the largest real solution x to (log x) log

x = T.

Relay 1-3 Let T = T N Y W R. Kay has T + 1 different colors of fingernail polish. Compute the number of ways
that Kay can paint the five fingernails on her left hand by using at least three colors and such that no two
consecutive fingernails have the same color.
Relay 2-1 Compute the number of ordered pairs (x, y) of positive integers satisfying x2 8x + y 2 + 4y = 5.
Relay 2-2 Let T = T N Y W R and let k = 21 + 2T . Compute the largest integer n such that 2n2 kn + 77 is a
positive prime number.
Relay 2-3 Let T = T N Y W R. In triangle ABC, BC = T and mB = 30 . Compute the number of integer values
of AC for which there are two possible values for side length AB.

25

10

Relay Answers

Answer 1-1

1
2

Answer 1-2 10
Answer 1-3 109890
Answer 2-1 4
Answer 2-2 12
Answer 2-3 5

26

11

Relay Solutions

Relay 1-1 A rectangular box has dimensions 8 10 12. Compute the fraction of the boxs volume that is not
within 1 unit of any of the boxs faces.
Solution: Let the box be defined by the product of the intervals on the x, y, and z axes as [0, 8][0, 10][0, 12]
with volume 8 10 12. The set of points inside the box that are not within 1 unit of any face is defined by
6810
the product of the intervals [1, 7] [1, 9] [1, 11] with volume 6 8 10. This volume is 81012
= 12 of the
whole box.
2

Relay 1-2 Let T = T N Y W R. Compute the largest real solution x to (log x) log
2

x = T.

1
2u

Solution: Let u = log x. Then the given equation can be rewritten as u

T = 0 2u2 u 2T = 0.

This quadratic has solutions u = 1 1+16T

. As we are looking for the largest real solution for x (and therefore,
4

for u), we want u = 1+ 1+16T

=
1
when
T
= 21 . Therefore, x = 101 = 10.
4
Relay 1-3 Let T = T N Y W R. Kay has T + 1 different colors of fingernail polish. Compute the number of ways
that Kay can paint the five fingernails on her left hand by using at least three colors and such that no two
consecutive fingernails have the same color.
Solution: There are T + 1 possible colors for the first nail. Each remaining nail may be any color except that
of the preceding nail, that is, there are T possible colors. Thus, using at least two colors, there are (T + 1)T 4
possible colorings. The problem requires that at least three colors be used, so we must subtract the number of
colorings that use only two colors. As before, there are T + 1 possible colors for the first nail and T colors for
the second. With only two colors, there are no remaining choices; the colors simply alternate. The answer is
therefore (T + 1)T 4 (T + 1)T , and with T = 10, this expression is equal to 110000 110 = 109890.
Relay 2-1 Compute the number of ordered pairs (x, y) of positive integers satisfying x2 8x + y 2 + 4y = 5.
Solution: Completing the square twice in x and y, we obtain the equivalent equation (x 4)2 + (y + 2)2 = 25,
which describes a circle centered at (4, 2) with radius 5. The lattice points on this circle are points 5 units
up, down, left, or right of the center, or points 3 units away on one axis and 4 units away on the other. Because
the center is below the x-axis, we know that y must increase by at least 2 units; x cannot decrease by 4 or
more units if it is to remain positive. Thus, we have:
(x, y)

(4, 2) + (3, 4) = (1, 2)

(x, y)

(4, 2) + (0, 5) = (4, 3)

(x, y)

(4, 2) + (3, 4) = (7, 2)

(x, y)

(4, 2) + (4, 3) = (8, 1).

There are 4 such ordered pairs.

Relay 2-2 Let T = T N Y W R and let k = 21 + 2T . Compute the largest value of n such that 2n2 kn + 77 is prime.
Solution: If k is positive, there are only four possible factorizations of 2n2 kn + 77 over the integers, namely
(2n 77)(n 1)

2n2 79n + 77

(2n 1)(n 77)

2n2 145n + 77

(2n 11)(n 7)

2n2 25n + 77

(2n 7)(n 11)

2n2 29n + 77.

Because T = 4, k = 29, and so the last factorization is the correct one. Because 2n 7 and n 11 are both
integers, in order for their product to be prime, one factor must equal 1 or 1, so n = 3, 4, 10, or 12. Checking

27

these possibilities from the greatest downward, n = 12 produces 17 1 = 17, which is prime. So the answer is
12.
Relay 2-3 Let T = T N Y W R. In triangle ABC, BC = T and mB = 30 . Compute the number of integer values
of AC for which there are two possible values for side length AB?
Solution: By the Law of Cosines, (AC)2 = T 2 + (AB)2 2T (AB) cos 30 (AB)2 2T cos 30 (AB) + (T 2
(AC)2 ) = 0. This quadratic in AB has two positive solutions when the discriminant and product of the roots
are both positive. Thus (2T cos 30 )2 4(T 2 (AC)2 ) > 0, and (T 2 (AC)2 ) > 0. The second inequality
implies that AC < T . The first inequality simplifies to 4(AC)2 T 2 > 0, so T /2 < AC. Since T = 12, we have
that 6 < AC < 12, giving five integral values for AC.

28

12

Tiebreaker Problems

Problem 1. In 4ABC, D is on AC so that BD is the angle bisector of B. Point E is on AB and CE intersects

AC
BD at P . Quadrilateral BCDE is cyclic, BP = 12 and P E = 4. Compute the ratio AE
.
Problem 2. Complete the following cross-number puzzle, where each Across answer represents a four-digit
number, and each Down answer represents a three-digit number. No answer begins with the digit 0. You
must write your answer in the blank 3 4 grid below.
Across:
1. A B C D is the cube of the sum of the digits in the answer to 1 Down.
5. From left to right, the digits in E F G H are strictly decreasing.
6. From left to right, the digits in I J K L are strictly decreasing.
Down:
1. A E I is a perfect fourth power.
2. B F J is a perfect square.
3. The digits in C G K form a geometric progression.
4. D H L has a two-digit prime factor.

Answer:
1

Problem 3. In rectangle M N P Q, point A lies on QN . Segments parallel to the rectangles sides are drawn
through point A, dividing the rectangle into four regions. The areas of regions I, II, and III are integers in
geometric progression. If the area of M N P Q is 2009, compute the maximum possible area of region I.

29

M
Y

B
A

M
A

III
Q

II

30

13

Tiebreaker Solutions

Problem 1. In 4ABC, D is on AC so that BD is the angle bisector of B. Point E is on AB and CE intersects

AC
BD at P . Quadrilateral BCDE is cyclic, BP = 12 and P E = 4. Compute the ratio AE
.
Answer 1. 3
Solution 1. Let denote the circle that circumscribes quadrilateral BCDE. Draw in line segment DE. Note
that DP E and CP B are congruent, and DEC and DBC are congruent, since they cut off the same arc
12
BC
= BP
of . Therefore, 4BCP and 4EDP are similar. Thus DE
EP = 4 = 3.
Because BCE and BDE cut off the same arc of , these angles are congruent. Let be the measure of
these angles. Similarly, DCE and DBE cut off the same arc of . Let be the measure of these angles.
Since BD is an angle bisector, mCBD = .
Note that mADE = 180 mBDE mBDC. It follows that
mADE

180 mBDE (180 mCBD mBCD)

mADE

180 mBDE (180 mCBD mBCE mDCE)

mADE

180 (180 )

mADE

2 = mCBD.

Thus ADE is congruent to CBD, and it follows that 4ADE is similar to 4ABC. Hence
AC
by substituting in given values, we have AE
= 3.

BC
DE

AC
AE ,

and

Problem 2. Complete the following cross-number puzzle, where each Across answer represents a four-digit
number, and each Down answer represents a three-digit number. No answer begins with the digit 0. You
must write your answer in the blank 3 4 grid below.
Across:
1. A B C D is the cube of the sum of the digits in the answer to 1 Down.
5. From left to right, the digits in E F G H are strictly decreasing.
6. From left to right, the digits in I J K L are strictly decreasing.
Down:
1. A E I is a perfect fourth power.
2. B F J is a perfect square.
3. The digits in C G K form a geometric progression.
4. D H L has a two-digit prime factor.

31

Answer 2.
1

2 1

Solution 2.

From 1 Down, A E I = 256 or 625, either of which make A B C D = 2197, so A E I = 256.

From B = 1 together with 2 Down, B F J = 121 or 144. But J = 1 does not work because then 6 Across could
not be satisfied. Therefore B F J = 144.
From C = 9 together with 5 Across and 3 Down, we have C G K = 931.
From D = 7 together with 5 and 6 Across, we get D H L = 720 or 710, but only 710 has a two-digit prime factor.
Problem 3. In rectangle M N P Q, point A lies on QN . Segments parallel to the rectangles sides are drawn
through point A, dividing the rectangle into four regions. The areas of regions I, II, and III are integers in
geometric progression. If the area of M N P Q is 2009, compute the maximum possible area of region I.

Answer 3. 1476
Solution 3. Because A is on diagonal N Q, rectangles N XAB and ACQY are similar. Thus
AB AY = AC AX. Therefore, we have 2009 = [I] + 2[II] + [III].

AB
AX

QY
QC

AC
AY

Let the common ratio of the geometric progression be pq , where p and q are relatively prime positive integers
(q may equal 1). Then [I] must be some integer multiple of q 2 , which we will call aq 2 . This gives [II] = apq
and [III] = ap2 . By factoring, we get
2009 = aq 2 + 2apq + ap2 72 41 = a(p + q)2 .
Thus we must have p + q = 7 and a = 41. Since [I] = aq 2 and p, q > 0, the area is maximized when
giving [I] = 41 36 = 1476. The areas of the other regions are 246, 246, and 41.

32

p
q

= 61 ,

14

Super-Relay

1. Quadrilateral ARM L is a kite with AR = RM = 5, AM = 8, and RL = 11. Compute AL.

2. Let T = T N Y W R. If xy = 5, yz = 5, and xz = T , compute the positive value of x.

3. Let T = T N Y W R. In how many ways can T boys and T + 1 girls be arranged in a row if all the girls must be
standing next to each other?
4. Let T = T N Y W R. Let T = T N Y W R. 4ABC is on a coordinate plane such that A = (3, 6), B = (T, 0), and
C = (2T 1, 1 T ). Let ` be the line containing the altitude to BC. Compute the y-intercept of `.
5. Let T = T N Y W R. In triangle ABC, AB = AC 2 = T , and mA = 60 . Compute BC 2 .
6. Let T = T N Y W R. Let S1 denote the arithmetic sequence 0, 41 , 12 , . . ., and let S2 denote the arithmetic sequence
0, 61 , 31 , . . . . Compute the T th smallest number that occurs in both sequences S1 and S2 .
7. Let T = T N Y W R. An integer n is randomly selected from the set {1, 2, 3, . . . , 2T }. Compute the probability
that the integer |n3 7n2 + 13n 6| is a prime number.
8. Let A be the number you will receive from position 7, and let B be the number you will receive from position
9. In A1 minutes, 20 frogs can eat 1800 flies. At this rate, in B1 minutes, how many flies will 15 frogs be able
to eat?
9. Let T = T N Y W R. If |T | 1 + 3i = z1 , compute the sum of the real and imaginary parts of z.

A
_

Let T = T N Y W R. In circle O, diagrammed at right, minor arc AB

10. measures T degrees. If mOAC = 10 and mOBD = 5 , compute the E
4
degree measure of AEB. Just pass the number without the units.

C
O
D
B

11. Let T = T N Y W R. Ann spends 80 seconds climbing up a T meter rope at a constant speed, and she spends
70 seconds climbing down the same rope at a constant speed (different from her upward speed). Ann begins
climbing up and down the rope repeatedly, and she does not pause after climbing the length of the rope. After
T minutes, how many meters will Ann have climbed in either direction?
12. Let T = T N Y W R. Simplify 2log4 T/2log16 64 .
13. Let T = T N Y W R. Let P (x) = x2 + T x + 800, and let r1 and r2 be the roots of P (x). The polynomial Q(x)
is quadratic, it has leading coefficient 1, and it has roots r1 + 1 and r2 + 1. Find the sum of the coefficients of
Q(x).
14. Let T = T N Y W R. Equilateral triangle ABC is given with side length T . Points D and E are the midpoints
of AB and AC, respectively. Point F lies in space such that 4DEF is equilateral and 4DEF lies in a plane
perpendicular to the plane containing 4ABC. Compute the volume of tetrahedron ABCF .
15. In triangle ABC, AB = 5, AC = 6, and tan BAC = 34 . Compute the area of 4ABC.

15

Super-Relay Answers

Answer to the relay: 3750

1. 4 5
2. 2
3. 36
1
9. 25
10. 5
11. 80

4. 3
12. 10

5. 19
13. 800

6. 9
7.
14. 108

33

1
9

8. 3750
15. 12

16

Super-Relay Solutions

2
2
1. Let K be the
midpoint of AM
. Then AK= KM = 8/2 = 4, RK = 5 4 = 3, and KL = 11 3 = 8.
Thus AL = AK 2 + KL2 = 42 + 82 = 4 5.

4
2. Multiply the three given
equations p
to obtain x2 y 2 z 2 = 5T
5. Thus xyz = 125T 2 , and the positive value

4
4
2
2
of x is x = xyz/yz = 125T /5 = T /5. With T = 4 5, we have x = 2.
3. First choose the position of the first girl, starting from the left. There are T + 1 possible positions, and then
the positions for the girls are all determined. There are (T + 1)! ways to arrange the girls, and there are T !
2
ways to arrange the boys, for a total of (T + 1) (T + 1)! T ! = ((T + 1)!) arrangements. With T = 2, the
answer is 36.

(1T )0
4. The slope of BC is (2T
1)T = 1, and since ` is perpendicular to BC, the slope of ` is 1. Because ` passes
through A = (3, 6), the equation of ` is y = x + 3, and its y-intercept is 3 (independent of T ).
5. By the Law of Cosines, BC 2 = AB 2 + AC 2 2 AB AC cos A = T 2 + (T + 2)2 2 T (T + 2) 21 = T 2 + 2T + 4.
With T = 3, the answer is 19.
6. S1 consists of all numbers of the form n4 , and S2 consists of all numbers of the form n6 , where n is a nonnegative
integer. Since gcd(4, 6) = 2, the numbers that are in both sequences are of the form n2 , and the T th smallest
such number is T 1
2 . With T = 19, the answer is 9.
7. Let P (n) = n3 7n2 + 13n 6, and note that P (n) = (n 2)(n2 5n + 3). Thus |P (n)| is prime if either
|n 2| = 1 and |n2 5n + 3| is prime or if |n2 5n + 3| = 1 and |n 2| is prime. Solving |n 2| = 1 gives n = 1
or 3, and solving |n2 5n + 3| = 1 gives n = 1 or 4 or 52 17 . Note that P (1) = 1, P (3) = 3, and P (4) = 2.
2
Thus |P (n)| is prime only when n is 3 or 4, and if T 2, then the desired probability is 2T
= T1 . With T = 9,
1
the answer is 9 .
s
15. Let s = sin BAC. Then s > 0 and 1s
= 43 , which gives s = 45 . The area of triangle ABC is therefore
2
1
4
1
2 AB AC sin BAC = 2 5 6 5 = 12.

14. The volume of tetrahedron ABCF is one-third the area of 4ABC times the distance
from F to 4ABC. Since
T 3
T
D and E are midpoints, DE = BC
=
,
and
the
distance
from
F
to
4ABC
is
2
2
4 . Thus the volume of

1 T2 3 T 3
T3
ABCF is 3 4 4 = 16 . With T = 12, the answer is 108.
13. Let Q(x) = x2 + Ax + B. Then A = (r1 + 1 + r2 + 1) and B = (r1 + 1)(r2 + 1). Thus the sum of the coefficients
of Q(x) is 1 + (r1 r2 2) + (r1 r2 + r1 + r2 + 1) = r1 r2 . Note that T = (r1 + r2 ) and 800 = r1 r2 , so the
answer is 800 (independent of T ). [Note: With T = 108, {r1 , r2 } = {8, 100}.]
1

1
12. Note that 2log4 T = 4( 2 log4 T ) = 4log4 T 2 = T . Letting log16 64 = x, we see that 24x = 26 , thus x = 23 , and
q

2x = 8. Thus the given expression equals T8 , and with T = 800, this is equal to 10.

T
11. In 150 seconds (or 2.5 minutes), Ann climbs up and down the entire rope. Thus in T minutes, she makes b 2.5
c
T
T
round trips, and therefore climbs 2T b 2.5 c meters. After making all her round trips, there are t = 60(T 2.5b 2.5 c)
t
seconds remaining. If t 80, then the

remaining distance climbed is T 80 meters, and if t > 80, then
t80
the distance climbed
 is T + T 70
 meters. In general, the totaldistance in meters that Ann climbs is
T
T
60
T
2.5b
c
60(T 2.5b 2.5
c)80
(
2.5 )
T
2T b 2.5
c + T min 1,
+ T max 0,
. With T = 10, Ann makes exactly 4
80
70

round trips, and therefore climbs a total of 4 2 10 = 80 meters.

34

10. Note that mAEB = 12 (mAB m CD) = 21 (mAB mCOD). Also note that mCOD = 360 (mAOC +
_

mBOD+mAOB) = 360 (180 2mOAC)(180 2mOBD)mAB = 2(mOAC+mOBD)mAB.

_

Thus mAEB = mAB mOAC mOBD =

9. Let t = |T |. Note that z =
of z is

t1
t2 2t+10

3
t2 2t+10

T
4

10 5 , and with T = 80, the answer is 5.

1
1
t13i
t13i
t1+3i = t1+3i t13i = t2 2t+10 . Thus the
|T |4
1
= |T |2 2|T |+10 . With T = 5, the answer is 25
.

sum of the real and imaginary parts

A
90 flies. Thus in
8. In A1 minutes, 1 frog can eat 1800/20 = 90 flies; thus in B1 minutes, 1 frog can eat B
A
1
minutes, 15 frogs can eat 15 90 B
flies. With A = 91 and B = 25
, this simplifies to 15 250 = 3750.

35

1
B